[ https://issues.apache.org/jira/browse/HADOOP-11362?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Steve Loughran updated HADOOP-11362: ------------------------------------ Attachment: 0001-HADOOP-11362-Test-org.apache.hadoop.crypto.random.Te.patch Patch 001 as a git format-patch formatted event. Culls the errant test method entirely. The other test method does what I'd proposed: repeated operations; all this does is log something and it is clearly unreliable at this, dependent on the underlying OS to work consistently > Test > org.apache.hadoop.crypto.random.TestOsSecureRandom.testOsSecureRandomSetConf > timing out > -------------------------------------------------------------------------------------------- > > Key: HADOOP-11362 > URL: https://issues.apache.org/jira/browse/HADOOP-11362 > Project: Hadoop Common > Issue Type: Bug > Components: test > Affects Versions: 3.0.0 > Environment: ASF Jenkins, Java 7 & 8 > Reporter: Steve Loughran > Attachments: > 0001-HADOOP-11362-Test-org.apache.hadoop.crypto.random.Te.patch > > > The test > {{org.apache.hadoop.crypto.random.TestOsSecureRandom.testOsSecureRandomSetConf}} > is timing out on jenkins + Java 8. > This is probably the exec() operation. It may be transient, it may be a java > 8 + shell problem. > do we actually need this test in its present form? If a test for file handle > leakage is really needed, attempting to create 64K instances of the OSRandom > object should do it without having to resort to some printing and manual > debugging of logs. -- This message was sent by Atlassian JIRA (v6.3.4#6332)